Policy Innovation Blog

Innovation policy through experimentation

Hasan Bakhshi - 27.04.2011

NESTA Provocations are independent essays by thinkers that showcase thought-provoking work on innovation. In our latest Provocation, State of Uncertainty: Innovation policy through experimentation, Jason Potts, Alan Freeman and I, writing in our personal capacities, call on the government to adopt a radically new approach to supporting innovation.

Our starting point is that uncertainty about future opportunities and constraints is the most important barrier to entrepreneurship. Innovation policy, we suggest, would work better if it were directed to the task of reducing this uncertainty. We challenge the idea, implicit in much existing practice, that the government operates levers that affect innovation in predictable ways. We argue that innovation policy should instead be conceived of as a process of discovery, required because the creation and exploitation of new ideas by entrepreneurs is by nature radically uncertain. There should be less emphasis on tax and spend, and more on experimental development, private-public clusters, demonstration projects and innovation funds.

Too much innovation policy remains rooted in industry policy, and has inherited many of its presumptions: such as a focus on planning, targets and sectoral programmes. This opens it to well-known critiques of industrial policy: that it makes heroic assumptions about the information government has access to, that planning structures will be captured,  that public bodies are inflexible in general, and that there is an absence of rigorous methods for evaluating its success or failure.

What might reform look like?

The guiding principle is that policy should enable experimental learning. That is, the goals of innovation policy need to be led by research and learning priorities. Where relevant knowledge is distributed across many agents, the government should use decentralised and networked structures to support entrepreneurs. Government should stand ready to scale up or down the experimental inquiry, perhaps significantly, depending on emerging findings. This would turn up and spread knowledge of business opportunities faster, and dramatically reduce the social cost of repetitive failure.

These arguments point to a more 'project-based' conception of innovation policy, which in turn means that support should be time-limited, with the routine inclusion of sunset clauses. This point may seem innocuous but it actually speaks to a significant institutional shift away from attempts to make some innovation policies into law (such as tax breaks for investors or long-term commitment to particular programmes). It instead seeks to internalise the notion that the policy ends when the learning ends - which may require policymakers to overcome opposition from established actors who have vested interests in maintaining policies forever.

In State of Uncertainty, we suggest that policymakers look at three particular models.

Strategic innovation funds

First, the idea of strategic 'innovation funds'. Large private or charitable funds are a common model for allocating resources in areas like science and social entrepreneurship. In a similar vein, public and private funds could be used to support collaborative business experiments which test propositions of wider business and/or social interest.

Testbeds

A second model would be for the state to provide 'testbed' environments for businesses, in which they can conduct and evaluate collaborative experiments. A related idea is the micro equivalent of 'special economic zones' or Paul Romer's notion of 'Charter Cities'. These would be special experimental zones that would offer, within a defined space, different rules or laws, or different bundles of public services. If effectively designed, councils or regions might be expected to compete to host these experiments, with the possible reward of an emerging cluster of new economic activity. Undertaking experiments in different settings is valuable because this helps us to learn about the conditions for success.

Studying success

A third model is a kind of business 'forensics laboratory' for the study of particular entrepreneurial, business and regional industrial successes and failures, seeking to discover what lessons they hold. This might be modelled on national science institutes, with a mandate to map the space of entrepreneurial opportunity and failure, much as observatories map, and make public, research about the physical universe. Economics departments in universities used to do a lot of this, but they don't do it so much now.

This new approach to innovation policy will require government to have different capabilities. In particular, it will need officials who have a deeper appreciation of research and evaluation, a culture more oriented towards learning and experimentation, and stronger entrepreneurial networks. Ours is a model of a learning state that sees its mandate in the gathering and dissemination of information that can be used in entrepreneurial action that drives innovation. The innovation goal of the government in this vision is to learn, not to produce.

As ever, I would welcome your comments.